A system and method for classifying unknown data patterns in a multi-variate feature space. An observer 12 evaluates the closeness of the data to any one of a plurality of known classes in a multi-variate feature space. A classifier 14 classifies the evaluated data into one of the plurality of known classes. A flagger 16 flags data having an unknown classification. A label requester 18 requests that the flagged data be interpreted into either a new class or one of the plurality of known classes. If the interpretation is a new fault, then a new class adder 22 sets up a new cluster representative of the fault and inputs the new class to the classifier 14. Alternatively, if the interpretation is that the flagged data belongs to one of the plurality of known classes, then a class resetter 24 resets the boundaries of this class and informs the classifier 14 of the change.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A system 26 for classifying data obtained from a process in a multi-variate feature space, comprising: an observer 12 that evaluates the closeness of the data to any one of a plurality of known classes defined in the multi-variate feature space; a classifier 14 for classifying the evaluated data into one of the plurality of known classes; a flagger 16 for flagging data having an unknown classification; a label requester 18 for requesting that the flagged data be interpreted into either a new class or one of the plurality of known classes; and a classifier adjuster 22 or 24 that adjusts the classifier according to the interpretation.
2. The system 26 according to claim 1 , further comprising a data scrubber 30 that scrubs the data.
3. The system 26 according to claim 1 , further comprising a preprocessor 32 that extracts features from the data.
4. The system 26 according to claim 3 , further comprising a normalizer that normalizes the data.
5. The system 26 according to claim 3 , wherein the observer 12 evaluates the closeness of the features to the plurality of known classes.
6. The system 26 according to claim 5 , wherein said flagger 16 issues a flag for features that are sufficiently far enough away from the plurality of known classes.
7. The system 26 according to claim 1 , wherein said label requester 18 requests that a user interpret the flagged data into either a new class or one of the plurality of known classes.
8. The system 26 according to claim 1 , wherein said classifier adjuster 22 or 24 comprises a new class adder 22 that adds a data cluster representative of the new class into said classifier 14 .
9. The system 26 according to claim 1 , wherein said classifier adjuster 22 or 24 comprises a class resetter 24 that resets the plurality of known classes in said classifier 14 to accommodate new class boundaries.
10. A method for classifying data obtained from a process in a multi-variate feature space, comprising: observing the closeness of the data to any one of a plurality of known classes defined in the multi-variate feature space; classifying the observed data into one of the plurality of known classes; flagging data having an unknown classification; requesting that the flagged data be interpreted into either a new class or one of the plurality of known classes; adjusting the classes in the multi-variate feature space according to the interpretation; and reclassifying the data according to the adjustment.
11. The method according to claim 10 , further comprising scrubbing the data.
12. The method according to claim 10 , further comprising extracting features from the data.
13. The method according to claim 12 , further comprising normalizing the data.
14. The method according to claim 12 , wherein the observing observes the closeness of the features to the plurality of known classes.
15. The method according to claim 14 , wherein said flagging issues a flag for features that are sufficiently far enough away from the plurality of known classes.
16. The method according to claim 10 , wherein said requesting requests that a user interpret the flagged data into either a new class or one of the at least one known classes.
17. The method according to claim 10 , wherein said adjusting comprises adding a data cluster representative of a new class.
18. The method according to claim 10 , wherein said adjusting comprises resetting the plurality of known classes to accommodate new class boundaries.
19. A multi-variate data assessment tool 10 for assessing data obtained from a process, comprising: an observer 12 that evaluates the closeness of the data to any one of a plurality of known classes defined in a multi-variate feature space for the process; a classifier 14 that classifies the evaluated data into one of the plurality of known classes; a flagger 16 for flagging data having an unknown classification; a label requester 18 for requesting that the flagged data be interpreted into either a new class or one of the plurality of known classes; a new class adder 22 that adds a data cluster representative of a new class into said classifier; and a class resetter 24 that resets the plurality of known classes in said classifier 14 to accommodate new class boundaries.
20. The tool 10 according to claim 19 , wherein said flagger 16 issues a flag for data that are sufficiently far enough away from the plurality of known classes.
21. The tool 10 according to claim 19 , wherein said label requester requests that a user interpret the flagged data into either a new class or one of the plurality of known classes.
22. A method for assessing data obtained from a process in a multi-variate feature space, comprising: extracting a plurality of features from the data; observing the closeness of each of the plurality of features to any one of a plurality of known classes defined in the multi-variate feature space; classifying the observed feature data into one of the plurality of known classes; flagging data having an unknown classification; requesting that the flagged data be interpreted into either a new class or one of the plurality of known classes; adjusting the plurality of known classes to accommodate the interpretation; and reevaluating the data in accordance with the adjusted classes in the multi-variate feature space.
23. The method according to claim 22 , further comprising scrubbing the data.
24. The method according to claim 22 , further comprising normalizing the data.
25. The method according to claim 22 , wherein said flagging issues a flag for the plurality of features that are sufficiently far enough away from the plurality of known classes.
26. The method according to claim 22 , wherein said requesting requests that a user interpret the flagged data into either a new class or one of the plurality of known classes.
27. The method according to claim 22 , wherein said adjusting comprises adding a data cluster representative of a new class.
28. The method according to claim 22 , wherein said adjusting comprises resetting the plurality of known classes to accommodate new class boundaries.
29. The method according to claim 22 , further comprising using a trend performance analysis tool to assess the data.
30. The method according to claim 29 , further comprising validating the reevaluated data with the trend performance analysis tool.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
April 24, 2000
February 11, 2003
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.